Pilot Testing the Feasibility of a Game Intervention Aimed at Improving Help Seeking and Coping Among Sexual and Gender Minority Youth: Protocol for a Randomized Controlled Trial

Introduction

Background

Sexual minority youth (eg, lesbian, gay, bisexual, or queer youth) and gender minority youth (ie, youth who identify as a gender different from their sex assigned at birth) experience myriad substance use and mental health disparities [1-27]. In particular, sexual orientation–related disparities have been known for over 20 years [1]: sexual minority youth, compared with heterosexual youth, have approximately 176% higher odds of cigarette use, 155% higher alcohol use, 34% higher heavy alcohol use (eg, binge drinking), and 56% higher marijuana use [2]. Currently, many of these sexual orientation–related disparities appear to be growing larger [3,4]. Similarly, compared with cisgender (ie, nontransgender) youth, gender minority youth have significantly higher use of cigarettes, alcohol, and marijuana [5,27]. These sexual and gender minority youth (SGMY) disparities are also present for novel substances, such as electronic cigarette use [5]. Regarding mental health disparities, SGMY have significantly higher anxiety, depression, and suicidality [20-27]. Meta-analyses show that sexual minority youth, compared with heterosexual youth, have 96% higher odds of having suicidal thoughts, 120% higher odds of making suicide plans, and 218% higher odds of making suicide attempts [20]. Altogether, these substantial and persistent health disparities make SGMY a priority population for interventions that attempt to reduce health inequities.

SGMY also experience disparities in bullying and cyberbullying victimization compared with their cisgender heterosexual peers [5,27-33]. For example, according to the 2015 Youth Risk Behavior Survey (YRBS), gay, lesbian, and bisexual adolescents (compared with heterosexuals) had nearly doubled the prevalence of bullying and cyberbullying victimization [33]. Importantly, research shows that these bullying disparities also contribute to SGMY disparities in substance use and mental health issues [5,13,27,34,35]. Therefore, interventions that help bullied SGMY and reduce bullying victimization may, in turn, reduce substance use and mental health disparities.

In addition to this greater prevalence of bullying, SGMY have unique factors that also contribute to health disparities. When SGMY are bullied, in addition to the typical fears of disclosing their bullying victimization experiences to others, they often fear having to disclose their sexual and gender minority (SGM) status to adults, thereby putting them at risk for further discrimination and harassment [36,37], likely preventing them from reaching out for help. Even when SGMY consider suicide, large proportions of them do not seek help [38], and sexual minority youth have more trouble than heterosexuals with identifying people to talk to about their emotional worries [39]. Moreover, compared with heterosexuals, sexual minority youth are more likely to use nonproductive coping strategies (eg, self-blaming, giving up, ignoring problems, and worrying) to manage the stressors in their lives [40,41], which likely exacerbates their health. Thus, interventions that aim to improve help-seeking behaviors and productive coping strategies (eg, solving problems, seeking relaxing diversions, and being physically active) among bullied SGMY may substantially reduce substance use and mental health disparities.

Notably, few interventions have been rigorously tested to examine whether they are efficacious in reducing substance use and mental health disparities among SGMY [42]. Moreover, 1 possible way to improve the health of bullied SGMY is via a Web-accessible (ie, downloadable via the internet) game intervention. Game interventions can be easily accessible through the internet, thereby providing an effective way to reach large numbers of SGMY, including SGMY living in rural areas and high structural stigma locations (ie, areas with less SGM-inclusive policies, institutions, and attitudes). Furthermore, Web-based game interventions are advantageous because some SGMY may be out to only a few people offline, making recruitment and attendance in face-to-face interventions for SGMY quite challenging [43,44]. Web-accessible interventions may be accessed by those who are insufficiently supported in offline programs [45-47]. In addition, Web-accessible gaming programs about sex [48,49], mental health [50], alcohol use [51-55], smoking [56-58], and asthma [59-62] are effective for youth in general. The Web-based environment is also relatively safe for LGB youth to gain coping skills [63,64]. Other advantages of Web-accessible interventions are increased fidelity and cost-effectiveness [65-71]. Overall, Web-accessible game interventions overcome common impediments of face-to-face interventions and present a unique opportunity to reach SGMY to improve their health. Little research, with few exceptions [72,73], has examined the feasibility of using a Web-accessible game intervention with SGMY populations. Therefore, pilot testing the feasibility of such a study is critical for successfully conducting a large-scale intervention. Feasibility comprises a wide range of topics [74], such as the testing of implementation procedures—how well the study was implemented as planned; intervention demand—the actual use of intervention among participants; intervention acceptability—how the participants react to the intervention; intervention integration—how well the game fits into the participants’ lives; intervention adaptation and expansion—the changes necessary for future iterations of the intervention and translation into new environments; and finally, limited efficacy testing—can be used to get estimates for variability and precision to power a larger randomized controlled trial (RCT). To adequately address the multiple dimensions of feasibility related to a game-based intervention tailored to SGMY, a carefully planned pilot trial is needed.

Study Aims

This paper describes the protocol for a pilot RCT assessing the feasibility and limited efficacy of the game-based intervention to increase help-seeking–related knowledge, intentions, self-efficacy and behaviors, productive coping skills use, and coping flexibility and reduce health risk factors and behaviors among SGMY. The primary, secondary, tertiary, and exploratory aims of the study are described below.

Primary Aim

Our primary aim is to evaluate implementation procedures for the RCT of the game-based intervention as well as to determine the level of game demand and game acceptability. We hypothesize having high implementation fidelity, game demand, and game acceptability (see Table 1 in the Methods section for specific targets).

Secondary Aim

Our secondary aim is to test the limited efficacy for the game in increasing the following short-term outcomes: help-seeking intentions, self-efficacy, and behaviors; productive coping strategies and coping flexibility; and knowledge and use of Web-based resources. We hypothesize that compared with the control participants, the intervention participants would have greater improvements in all short-term outcomes.

Tertiary Aim

Our tertiary aim is to test the limited efficacy for the game in reducing long-term outcomes: bullying and cyberbullying victimization; loneliness; mental health issues (ie, stress, anxiety, depression, and suicidality); substance use; and internalized SGM stigma. We hypothesize that intervention participants, compared with control participants, will have greater decreases in all long-term outcomes.

Exploratory Aim

We have built in an exploratory aim meant to better our understanding of participants’ responses to the game and research procedures and how to improve both. This exploratory aim concerns implementation procedures, integration, and the adaptation and expansion of the game. Given the exploratory nature of this aim, we have no a priori hypotheses.

Methods

Study Design

The purpose of this research study is to pilot a 2-arm RCT of a game-based intervention to improve help-seeking behaviors and productive coping strategies to reduce substance use, victimization, and mental health issues among SGMY. The study is led by a team with expertise in SGMY research at the Center for LGBT Health Research in the University of Pittsburgh’s Graduate School for Public Health, University of Pittsburgh (Principal Investigators: Friedman and Egan). We engage in team science by partnering with an expert in bullying research, help-seeking research, an interventionist, a biostatistician, a health and game researcher, and a professional game development company. The study is funded by the Eunice Kennedy Shriver National Institute of Child Health and Human Development at the National Institutes of Health (R21HD083561) and is registered as a clinical trial (NCT03501264) [75].

Study Population and Study Flow

All study procedures related to screening, consenting, surveying, and reminding are completed using REDCap, a free and secure Health Insurance Portability and Accountability Act-compliant website for managing Web-based surveys and databases. Study participant flow is depicted in Figure 1.

First, to determine eligibility, respondents completed a brief Web-based self-reported screening questionnaire before entering in this study. Respondents were eligible if they were English literate, lived in the United States of America, were aged between 14 to 18 years, had experienced bullying or cyberbullying victimization in the past year, had a sexual minority identity (ie, gay, lesbian, bisexual, or queer) or a gender minority identity (ie, considered themselves to be transgender or nonbinary), had a personal computer or Mac laptop or desktop computer where they could download games, and had an email address. Eligible respondents were then directed to a Web-based informed consent form. Participants voluntarily consented using a click-to-consent procedure. To protect participants from having to reveal their sexual or gender minority identities to their caregivers, thereby potentially putting them in harm’s way, we received a waiver of parental consent [76]. This allowed participants to self-consent. Ineligible respondents were thanked for their time and no additional contact was made.

Once eligible respondents agreed to voluntarily participate in the study, they were emailed a link to the baseline (T1) survey to complete. The T1 survey contained 24 pages with a median of 8.5 items per page (mean 10.3; range 3-26). After completion of the T1 survey, the study team randomized participants to intervention or control conditions (full descriptions of conditions are provided in later sections). Immediately after randomization, participants in the intervention condition automatically received an email with a REDCap survey link containing information that guided participants through the procedures for downloading and installing the game intervention onto their desktop or laptop computers with Microsoft Windows or Apple Mac operating systems. Every 3 days thereafter, participants were automatically sent up to 5 email reminders to download the game. Moreover, 1 day after randomization, participants in both the intervention and control conditions were automatically sent a list of resources related to study outcomes.

The first follow-up (T2) survey was activated 4 weeks after T1 survey completion and remained open for 4 weeks. The T2 survey was similar to the T1 survey; however, participants in the intervention condition who self-reported having played the game also completed questions about their gaming experience. The T2 survey contained 26 pages with a median of 8.5 items per page (mean 11.4; range 3-47).

The final follow-up survey (T3) is activated 8 weeks after T1 survey completion and remains open for 8 weeks. The T3 survey is similar to the T1 survey; however, intervention condition participants who do not complete the gaming experience questions at T2 but self-report having played the game in the T3 survey will be asked questions about their gaming experience at T3. The T3 survey contained 26 pages with a median of 8.5 items per page (mean 11.4; range 3-47).

For each survey, up to 5 email reminders to complete surveys are automatically emailed to participants every 4 days. While completing each survey, participants were able to change their answers by clicking a back button. The following incentives are given after the completion of each survey: US $10 for T1; US $25 for T2; and US $50 for T3. At the end of each survey, participants select if they wanted a gift card to Apple iTunes or Google Play.

Recruitment

Participants were conveniently sampled and recruited throughout the United States using website advertisements posted on social media platforms. This passive approach allowed SGMY from multiple geographic locations (eg, rural and urban, and East and West) to enroll in the study without overextending our limited resources. Facebook was our primary recruitment site. Facebook is an appropriate recruitment platform because it is highly utilized by adolescents; approximately 71% of teens use Facebook [77]. We created a formal side-bar Facebook ad (in-line ads were not used). We also recruited participants from Instagram using the same advertisements. We also recruited participants from SGM-related Web-based gaming groups, such as Geeks OUT, Gay Geeks (Facebook group), GaymerX (Facebook group), Transmission Gaming, and Reddit Gaymer forums. We advertised via Pitt+Me, a community hosted by the University of Pittsburgh comprising patients, volunteers, and researchers working together as partners in research and clinical trials to advance health care. Finally, we advertised on a dedicated Facebook page for the study. Only study team members were able to post comments on this page. An individual who was interested in participating clicked on the advertisement and was directed to the Web-based screening questionnaire. To ensure representation of both sexual and gender minorities, enrollment was monitored weekly. Specific ads were created and used to target underrepresented groups. Depending on the prior week’s enrollment numbers, we tailored which ads were used for the upcoming week.

Randomization

We used permuted block allocation (using blocks of several sizes) to randomize individual participants to the intervention or control conditions. The permuted blocks were created using the ralloc package for Stata. Randomization was performed in REDCap using the Randomization Module.

Control Materials

All participants randomized into the control condition received a list of national SGM-inclusive resources. This list included general lesbian, gay, bisexual, transgender, and queer (LGBTQ) resources (eg, GLAAD resource list), LGBTQ bullying resources (eg, The Trevor Project), general bullying victimization resources (eg, StopBullying.gov), child abuse resources (eg, Child Abuse Resource Center), dating violence resources (eg, National Domestic Violence Hotline), suicide and mental health resources (National Suicide Prevention Lifeline), substance use resources (eg, National Institute of Drug Abuse for Teens), LGBTQ homelessness (eg, True Colors Fund), and LGBTQ emergency hotlines (eg, Gay, Lesbian, Bisexual, and Transgender National Help Center Hotline). These materials were delivered via email the day after participants were randomized to the control condition. In addition, after completion of the final T3 survey, control participants were offered a free download of the intervention game. No additional follow-up was initiated to assess their game use, game satisfaction, or changes in outcomes.

Intervention Materials

Participants assigned to the intervention condition received a list of SGM-inclusive resources (the same as the control condition materials) and the game intervention. After randomization, intervention condition participants received an email with a link and instructions on how to download the game. Participants were asked to download the game to their computer to play it. The game intervention was based on empirical interviews with SGMY about their gaming preferences, undergirded by etiologic and behavioral change theories and built in collaboration with expert educational game developers.

Gaming Preferences of Sexual and Gender Minority Youth

Before developing the game, we conducted one-on-one, in-depth interviews with 20 SGMY about their gaming preferences (publication of these results is currently underway). Most SGMY enjoyed playing action-oriented games, and when asked what they liked about their favorite games, the most common response concerned the ability to personalize characters. Therefore, in the game, we incorporated selection of pronouns (Figure 2) and character customization, including skin color, hair, body type, and clothing (Figure 3). SGMY liked engaging, unique storylines, as well as challenging (but not too challenging) mini objectives and missions. Having a multiplayer aspect of the game that requires teamwork and interaction with others was also mentioned. We also incorporated these elements into the game.

‎View this figure

‎View this figure

‎View this figure

Theoretical Underpinnings

The conceptual model of the game intervention (Figure 4) is based on 3 theories that inform the etiology and behavioral change strategies of our short-term and long-term outcomes. First, social cognitive theory [78,79] suggests that behavior change (eg, help-seeking behaviors) is facilitated by developing self-efficacy and skills. Self-efficacy and social skills can be achieved through behavioral rehearsal, witnessing outcomes of their choices, and feedback [80-82]. Second, stress and coping theory [83] suggests that specific types of appraisals predict different coping strategies. With bullying, youth who blame themselves, perceive little control, and view the situation primarily as a threat as opposed to a challenge are likely to engage in nonproductive coping [84-87]. On the other hand, identification of the problem, direct problem solving, and help seeking constitutes productive coping [88]. Third, the social and emotional learning framework [89] identifies 4 main competencies to promote positive health outcomes: awareness of self and others, responsible decision making, positive attitudes and values, and social interaction skills. We considered each of these theories during the process of developing the core game mechanics of the player experience.

Elements of these theories are embedded in the game in unique ways. First, the game encourages help-seeking behaviors by having players create a team with other nonplayable characters and to pair a lonely nonplayable character with an appropriate mentor (Figure 5). Second, the game encourages use of productive coping strategies through active listening and helping a nonplayable character overcome anger in a healthy way (Figure 6). Third, the game raises awareness of Web-based resources through collecting pages from a virtual notebook that contain information about external resources and bullying information (Figures 7 and 8).

‎View this figure

‎View this figure

‎View this figure

‎View this figure

‎View this figure

Game Development

The game was developed in collaboration with Schell Games, an education and entertainment game development company located in Pittsburgh, Pennsylvania, United States. The intervention game is a role-playing game inspired by Japanese role-playing games, which typically involve exploring a space, talking to nonplayable characters, and fighting enemies through turn-based battles. The game runs standalone on desktops and laptops with Windows or Mac operating systems. Before implementation, we user-tested the intervention using think-aloud interviews with 3 SGMY, who found technical bugs and recommended password protection. Prior research shows that user testing with 3 participants results in finding at least 60% of major usability problems [90]. Schell Games revised the game accordingly before the implementation of the RCT.

Game Play

In the game, the player takes on the role of a Singular, a superhuman individual with special gifts, who is located in a school. The player is told that because of their uniqueness, Singulars face prejudice, often driven by fear and misunderstanding. After customizing their character, the player is tasked with finding a team to help them complete their final mission, which is to become a world-class superhero by defeating the robots in the Holochamber Challenge. The player then explores the school, talking to potential peers (ie, nonplayable characters) who can join their team. Some peers deal with problems related to bullying, confidence issues, and anger, which prevent them from performing properly or from wanting to join the player’s team. The player’s mini objectives are to do the following for each of their peers: (1) best identify the nonplayable characters’ problems, (2) find the best individuals or resources to help the nonplayable characters, and (3) help the nonplayable characters properly communicate or utilize their newfound resources. If the player is successful and finds the best way to help the nonplayable characters, then the nonplayable characters will join their team or help them by giving them an item or ability that will help them achieve success in the final Holochamber Challenge (Figure 9). After the player finishes the Holochamber Challenge, the events that took place are evaluated. For every nonplayable character that is successfully helped, the player is given a positive ending; for example, “Thanks to Invisibella’s mentoring, Violet Phantom eventually felt comfortable to reveal herself as a Singular. Despite past events, a majority of her other friends and family were supportive. Knowing she was not alone and had support helped her no longer feel afraid.” For every nonplayable character that was not successfully helped, the player is given a negative ending; for example, “Gigaton didn't learn to control his anger in this round. One day he snapped and turned on those bullying him. Though no one was seriously injured, the outburst resulted in Gigaton getting in trouble with the Principal.” The players are then encouraged to replay the game and given hints to help them receive positive endings. Each player can unlimitedly replay the game.

Data Collection

We are collecting data in 2 distinct ways. First, we use self-administered surveys via REDCap, where participants completed surveys on their computer, tablet, or phone. Participants complete surveys at T1, T2, and T3 (see the section Study Population and Study Flow for more details). Surveys were created based on the Checklist for Reporting Results of Internet E-Surveys [91]. Second, we collect game play data from participants in the intervention condition. These game play data are transferred via a secure file-transfer-protocol system. Text files containing milestones achieved, time played, and player choices are automatically sent to a secure server housed at the University of Pittsburgh. Participants’ game play data are tracked using a unique identification number that does not rely on identifiable information (eg, IP addresses).

Outcomes

Primary Outcomes

To answer our primary aim, we will assess the following primary feasibility outcomes: success of the implementation procedures used in our RCT, game demand, and game acceptability. Table 1 details our primary outcomes, assessments, and investigator-generated hypotheses.

Success of our implementation procedures are assessed using a variety of measures. We hypothesize the following: 240 participants will enroll in our trial and complete the first survey, greater than or equal to 80% of eligible participants will consent to participate in the study, 120 participants will be randomized to the intervention condition and 120 to the control condition, greater than or equal to 80% of enrolled participants will complete the T2 survey, greater than or equal to 80% of enrolled participants will complete the T3 survey, and greater than or equal to 75% of participants will complete all surveys. Testing the feasibility of our implementation procedures is essential to inform the design of our future larger-scale RCT.

‎View this figure

Game demand is assessed among the intervention condition participants in 2 ways, via downloading and playing the game. Download of the game is assessed via self-reported surveys (T2 and T3). We hypothesize ≥80% of intervention condition participants will download the game. Game play is assessed via self-reported surveys (T2 and T3) and game play data. We hypothesize ≥80% of intervention condition participants will play the game and ≥75% of intervention participants will play the game for 1 hour or greater.

Game acceptability among the intervention condition participants is assessed via self-reported surveys at T2 and T3. If participants reported playing the game, we ask about their overall impressions of the game using the Gaming Experience Questionnaire. The Gaming Experience Questionnaire is a multidimensional scale that assesses the following domains about the participants’ feelings and thoughts during the game: competence, sensory and imaginative immersion, game flow, game tension and annoyance (reverse coded), negative affect (reverse coded), and positive affect. We hypothesize that the mean score for each subscale will be greater than 2, representing a moderately good gaming experience. In addition, we ask participants if they would be interested in playing the game again, and whether they would recommend the game to their friends. We hypothesize ≥75% of participants will “agree” or “strongly agree” that they would like to play the game again, and ≥75% would “definitely,” “very probably,” or “probably” recommend the game to their friends.

Secondary Outcomes

To answer our secondary aim, we will measure the following short-term outcomes: help-seeking intentions, self-efficacy, and behaviors; productive coping strategy usage and coping flexibility; and knowledge and use of Web-based resources. All secondary outcomes are assessed via self-reported surveys at all time points. Table 2 details our secondary outcomes, items, response options, and coding procedures.

Help-seeking intentions are assessed using adapted version of the General Help Seeking Questionnaire [93]. In addition, 2 different sets of questions assess how likely participants are to seek help from a variety of sources about (1) emotional problems and (2) suicidal ideation. Help sources include a wide variety of people (eg, doctor and counselor) and places (eg, websites).

We adapted the original items by adding “Phone or text/chat help line (such as the National Suicide Prevention Lifeline or the Trevor Project)” and “teacher.” We will create 2 separate overall mean scores for how likely participants are to reach out concerning (1) emotional problems and (2) suicidal ideation. In addition, we will create a score for each help source individually. Higher mean scores indicate greater likelihood of seeking help from these sources.

Help-seeking self-efficacy is assessed using 2 subscales from Bandura’s Multidimensional Scales of Perceived Self Efficacy [94,95]: the enlisting social resources subscale and the enlisting parental and community support subscale. Each subscale has 4 items, which are used to calculate an overall mean score measuring how well participants think they can get support from different sources (eg, “How well can you get a friend to help you when you have social problems” and “How well can you get your parents to take part in school activities”).

Help-seeking behaviors are assessed using items adapted from the Help-Seeking Behaviors Scale [96]. We adapted the scale to include the wide variety of help sources used in the aforementioned help-seeking intention scales. Therefore, we created 8 different subscales (1 for each help source) assessed with 3 items each (eg, “In the past month, how often have you asked a teacher for help?”). Each subscale will be analyzed separately.

Coping skill usage is assessed using the Adolescent Coping Scale Second Edition Short Form [88]. This scale uses 18 items measuring 2 different dimensions: productive (problem solving) coping and nonproductive (passive avoidant) coping. We hypothesize that the game-based intervention (vs the control) will increase productive coping and decrease nonproductive coping.

Coping flexibility is assessed using the Coping Flexibility Scale [97], including 2 separate subscales: evaluation coping and adaptive coping. The evaluation coping subscale uses 5 items to assess how well a person monitors and evaluates coping outcomes (eg, “I am aware of how successful or unsuccessful my attempts to cope with stress have been.”). The adaptive coping subscale uses 5 items to assess how well a person uses an alternative coping strategy to produce a desirable outcome (eg, “When a stressful situation has not improved, I try to think of other ways to cope with it”). Each subscale will be examined separately.

Knowledge and use of Web-based resources are measured using 2 different scales developed by the investigative team. We created a list of SGM-inclusive resources that match what was provided in the game-based intervention and control materials, such as the Trevor Project. Use of Web-based resources is measured for the past month. We will create 1 summary score for knowledge and 1 for use; each summary score adds together all “yes” response options.

Tertiary Outcomes

To answer our tertiary aim, we will measure the following long-term outcomes: bullying and cyberbullying victimization, loneliness, mental health issues, substance use, and internalized SGM stigma. All tertiary outcomes are assessed via self-reported surveys at T1, T2, and T3. Table 3 details our tertiary outcomes, items, response options, and coding procedures.

Bullying victimization is measured using 2 different scales, 1 for traditional face-to-face bullying and 1 for cyberbullying. We assess past-month bullying victimization using an adapted version of the University of Illinois Victimization Scale [98,99], which consists of 6 items assessing the number of times respondents experienced harassment in the past month. In addition, 4 response options range from “never” to “7 or more.” These response options are assigned a value from 1 to 4, and a mean score is calculated, with a higher mean score indicating greater bullying victimization. We assess past-month cyberbullying victimization using 4 items adapted from an internally consistent cyberbullying perpetration scale [100]. Response options use a 5-point Likert scale including “not sure,” “never,” “rarely,” “occasionally,” and “often,” which we will recode to a 4-point scale, where “never” equals 1 and “often” equals 4. We will average scores across all items, with a higher mean score indicating greater cyberbullying victimization.

Loneliness is measured via Robert’s Version of the University of California, Los Angeles Loneliness Scale [101], a validated measure of loneliness. Overall, 8 items (eg, “I feel isolated from others”) are used to calculate a mean score for how lonely participants felt.

Mental health issues include stress, anxiety, depression, and suicidality. Past-month stress is assessed using the Perceived Stress Scale [102,103], which contains 10 items used to calculate a mean score. Past-week anxiety symptoms are assessed using the Severity Measure for Generalized Anxiety Disorder—Child Age 11-17 [104,105], which contains 10 items used to calculate a mean score. Past-week depressive symptoms are assessed using the Patient Health Questionnaire-9 for children aged from 11 to 17 years [106], which contains 9 items used to calculate a mean score. Suicidality is assessed with 3 questions adapted from the 2017 YRBS with items measuring suicidal thoughts, plans, and attempts. Although the YRBS questions measured past year suicidality, we adapted them to assess for suicidality in the past month. Each of these 3 suicidality items will be modeled separately as dichotomous (presence vs absence) variables.

Substance use includes alcohol, cigarette, electronic cigarette, and marijuana use. All items are assessed based on the YRBS and adapted to measure past-month use. Alcohol use is assessed via 2 questions: 1 measuring the frequency of past-month alcohol use and 1 item assessing frequency of binge drinking (having 5 or more drinks in a row). Similar to prior research [7], each item will be modeled as continuous variable representing the number of days used. Cigarette smoking is assessed via 2 items measuring past-month number of days smoked and average quantity smoked per day. Cigarette smoking will be modeled continuously by multiplying smoking frequency by quantity, similar to prior research [10]. Electronic cigarette smoking and marijuana use are both assessed using single items that measure past-month frequency of use. Each will be modeled as a continuous variable representing the number of days used. For all substance use items, we will use midpoints for categories with a range (eg, “1 or 2 days” will be coded as 1.5).

Internalized SGM stigma is measured using 2 different scales. To assess internalized gender minority stigma, we use the Transgender Identity Survey [107,108], which consists of 26 items that are divided into 4 different subscales: pride (reverse coded), passing, alienation, and shame. We adapted this scale by adding the term “nonbinary” to each item. To assess internalized sexual minority stigma, we adapted the Transgender Identity Survey [107,108] items by replacing “transgender or nonbinary” with “lesbian, gay, bisexual, or queer”; otherwise, the subscales, response options, and scoring are identical.

Exploratory Outcomes

To answer our exploratory aim, we assess several outcomes meant to better our understanding of participants’ responses to the game and how to improve the game in the future. Our exploratory aim concerns implementation procedures, intervention integration, and intervention adaptation and expansion outcomes. Table 4 details our exploratory outcomes, research questions, and assessments.

Multiple implementation procedures will be explored to inform our future research. These include the following: how long it takes to enroll participants on the internet, which venues participants were recruited from, how long it takes participants to complete the surveys, how many participants complete the game, and which game milestones are achieved by participants. These data are obtained from multiple sources, including the self-reported surveys in REDCap and the game play data.

Implementation integration—or how well the game fits into the participants’ lives—will be assessed in a variety of ways. We will examine how many participants were excluded from our study based on not having access to a computer or not having an email address. We will examine if the participants had any problems downloading or playing game and how easy it was for them to participate in the RCT. We will also explore if participants had issues with game security, privacy, or interference. These outcomes are tracked using participants’ responses to the screening questionnaire, T2 survey, and T3 surveys, as well as from participants’ correspondences with the project email address.

Adaptation and expansion of the game intervention will be assessed via the T2 and T3 surveys. We will ask participants if they prefer to play the game on a different format (eg, on their phones), their favorite and least favorite parts of the game, how they would change and improve the game, and whether they think other SGMY would be interested in the game. Finally, we will explore whether the responses to the Gaming Experience Questionnaire [92] subscale scores differed by participants’ gender, sexual orientation, race, ethnicity, and age.

Demographics and Potential Confounders

Demographic variables include age, grade in school, race, ethnicity, parent’s highest education level, and eligibility for free or reduced-price lunch at school. We also measure gender using the 2-step process assessing current gender identity and sex assigned at birth [112]. Moreover, 2 items assess gender expression. We measure sexual orientation using 3 questions assessing sexual attraction, behavior, and identity.

In addition to demographic variables, we assess variables that may confound the relationship between the intervention and our outcomes. Structural stigma is assessed by collecting participant’s residential zip code. Each zip code will then be given a state-level structural stigma score using Hatzenbuehler scale [113], based on density of same-sex couples, inclusive policies (eg, employee nondiscrimination policies), public opinion, and percentage of high schools with gay-straight alliance (also known as a gender and sexuality alliance; GSA). We have participants self-report whether or not their school had a GSA. We assess the participants’ overall school environment with 5 questions from the California School Climate Survey. We assess participants’ level of perceived social support using the Multidimensional Scale of Perceived Social Support, measuring support from family and friends [114]. Finally, we measure level of “outness” regarding participant’s sexual orientation and gender identity.

Statistical Analysis

For the measures of implementation procedures, demand, acceptability, integration, and adaptation and expansion (ie, primary and exploratory outcomes), we will report results using descriptive statistics (ie, percentages and frequencies for categorical variables or means and SDs for continuous variables) along with 95% CIs. We will compare our results against our a priori benchmarks (see Table 1). For the limited-efficacy outcomes (ie, secondary and tertiary outcomes), we will use repeated measures (ie, multilevel) statistical models using linear or logistic regression, depending on the distribution of the outcomes. To examine whether there are greater improvements over time in the intervention versus control conditions, we will test the interaction term of time by condition, wherein we will focus on the point estimates and 95% CIs. We will conduct our primary analyses as an intent-to-treat analysis, wherein everyone in the intervention condition is treated equally. As an additional exploratory exercise, we will conduct an analysis wherein we assess whether intervention effects were stronger with more intensive uptake of the game intervention. For these analyses, instead of a binary intervention indicator variable (1.0 for intervention vs 0.0 for control), we will use an intensity-adjusted intervention variable. The intensity-adjusted intervention variable will be coded as 1.0 for those who complete the game, 0.5 for those who play the game, and 0.0 for those who never played the game or were in the control condition. All analyses will be completed in Stata 15.0 (StataCorp LLC, College Station, Texas), and significance will be set as at P<.05.

Qualitative Analysis

For the open-ended questions, we will conduct qualitative data analysis. First, a group of investigators will read all of the participants’ responses and inductively identify the main themes for each item. Second, using these themes, we will develop a codebook with code names and definitions. Finally, 2 independent coders will code the qualitative data using Dedoose. If the 2 coders cannot reach consensus, a third coder will resolve any disagreements.

Sample Size and Power Calculation

This is a feasibility study, which is primarily being conducted to inform a larger RCT. As such, we powered this study based on our primary outcomes [74,115]. For our primary outcomes regarding the success of implementation procedures (eg, retention rate at T2), assuming 240 participants and 5% type 1 error rate, we will be able to estimate 95% CI width of no more than 0.13. For game demand among 120 intervention condition participants, we will be able to estimate 95% CI width of no more than 0.18. For our secondary and tertiary outcomes, we are primarily interested in effect size and CI estimation; we are not necessarily interested in finding statistically significant effects [74,115]. Estimations of the effect size and CI width will help us power a future, larger RCT.

Ethics Statement

All study procedures were approved by the Human Research Protection Office at the University of Pittsburgh. To protect participants from having to reveal their sexual or gender minority identities to their caregivers, thereby potentially putting them in harm’s way, we received a waiver of parental consent [76]. This allowed participants to self-consent. We did not collect personally identifying information other than email, phone, and zip code. Incentives, in the form of either Apple or Google electronic gift cards, are sent via email. This study was covered by a Certificate of Confidentiality from the National Institutes of Health.

Results

Enrollment and Randomization

Overall, 2153 individuals clicked the link to the screening questionnaire (Figure 10). In total, 988 individuals completed the screening questionnaire, of which 407 individuals met all eligibility criteria. Overall, 304 individuals consented to participate, and 240 participants completed the T1 survey and were randomized. Of those who were eligible, 59.0% (240/407) were enrolled into the study. Half of the enrolled participants (n=120) were randomized into the intervention condition and half (n=120) into the control condition. All 240 participants were enrolled into the RCT between April 2018 and July 2018. The final surveys (T3) will be completed in November 2018.

Sample Demographics

At baseline, there were no significant differences in demographic characteristics between intervention and control condition participants (Table 5). Overall, 52.1% (125/240) of participants identified as gay or lesbian, 26.7% (64/240) as bisexual, 24.2% (58/240) as queer, 11.7% (28/240) as another nonheterosexual identity, 4.6% (11/240) as unsure, and 2.1% (5/240) as heterosexual (all of whom were gender minorities). Nearly half (47.1%; 113/240) of participants were a gender minority, 36.7% (88/240) were cisgender boys, and 16.3% (39/240) were cisgender girls. Among the 113 gender minority participants, 54.9% (62/113) identified as a transgender boy, 33.6% (38/113) as nonbinary, 18.6% (21/113) as a boy, 17.7% (20/113) as genderqueer, 6.2% (7/113) as a transgender girl, 5.3% (6/113) as a girl, and 5.3% (6/113) as another identity (data not shown; participants could select multiple options). Overall, 81.3% (195/240) of participants identified as white, 20.4% (49/240) as Hispanic/Latinx, 10.8% (26/240) as black or African American, 7.5% (18/240) as American Indian or Alaska Native, and 5.8% (14/240) as Asian. Moreover, 36.7% (88/240) of participants were eligible for free or reduced-price lunch, and 52.9% (127/240) had a parent/guardian with a college degree. Figure 11 shows the geographic distribution of participants across the United States of America. In total, participants lived in 30 states.

‎View this figure

‎View this figure

Discussion

Principal Findings

In this pilot RCT, a game-based intervention, we successfully enrolled and randomized 240 SGMY participants over a 4-month period. Recruitment was most successful through Facebook and Instagram. Over half of the enrolled sample identified as gay or lesbian and nearly half were gender minority youth. There were no demographic differences between intervention and control conditions.

This study protocol offers several strengths and novel methods that can inform future intervention studies with SGMY. Our game intervention is one of the first Web-accessible programs targeting SGMY who are victims of bullying or cyberbullying. Furthermore, this study engages a younger cohort of individuals focusing on youth aged 14 to 18 years, as opposed to the typical focus of SGMY-relevant intervention research, which concentrates on those aged 18 years and older. This is also one of the first bullying prevention programs to address social and emotional learning (eg, coping strategies) with respect to bullying among SGMY utilizing a gaming format to attract participation and build retention. Finally, the internet-based distribution of the game intervention has the potential to reach far greater numbers of SGMY than traditional face-to-face interventions.

Limitations

Despite the many strengths of this study protocol, it is not without limitations. Of particular concern is selection bias. Participants in our study were conveniently sampled via the internet. Participants were also required to have a computer with internet access and an email address. Therefore, our study results may not be generalizable to all SGMY (eg, those who are homelessness or have a low socioeconomic status). In addition, participants can only access the game intervention on desktop or laptop computers with Windows or Mac operating systems. In future iterations, we plan to make the game intervention compatible with smartphones, thereby increasing the accessibility and potential usability of the game. Although we used and adapted measures with strong psychometric properties, most of the scales were not validated specifically among SGMY. Finally, an 8-week follow-up period may be too brief to affect our identified long-term outcomes; however, we chose to measure these outcomes as they are important health outcomes to inform a longer-term intervention study.

Conclusions

SGMY experience great disparities in bullying victimization, substance use, and mental health; however, there are few scientifically tested interventions currently available to reduce these disparities. This paper describes our protocol for the study of a Web-accessible game intervention aimed at improving the health of SGMY. Specifically, the outcomes of this pilot study are to assess the feasibility of implementing a game intervention to SGMY and limited efficacy of this intervention: increasing help-seeking–related knowledge, intentions, self-efficacy, and behaviors; increasing productive coping skills use and coping flexibility; and reducing health risk factors and behaviors among SGMY. Our study protocol directly informs the scientific development of a future, larger RCT testing the limited efficacy of our game intervention.